Chapter 4: Chapter 4: Exploration of Phenothiazine Derivatives as Singlet Oxygen-Mediated Photoremovable Protecting Groups Photoremovable Protecting Groups
4.2 Introduction
4.2.1 Photoremovable Protecting Groups and Singlet Oxygen-Mediated Deprotection
Protecting groups are integral in both chemistry and biology. In chemistry, they help define reactivity and selectivity, affording new approaches for complex syntheses. In biology, they allow for the regulation of complex networks. Although the protecting groups used in the two disciplines vary drastically, they tend to overlap in the context of photochemistry. Photoremovable protecting groups have been shown to be immensely impactful in chemical biology since the first reports of photochemically caged cAMP and ATP were published approximately 50 years ago.1,2
Over the past 50 years, photochemists have primarily focused on trying to push the speed, selectivity, and applicability of photochemical processes, irrespective of the wavelength they utilize.3 Recently,
however, an interest in in vivo biological applications for photoremovable protecting groups has brought to light the need for photochemical reactions that occur at longer wavelengths, where greater tissue penetration is obtained and less biological damage is incurred.3 Significant progress has been made in this area – as a number of new reactions have been developed that utilize visible light – however, there is still much room for improvement.4–9
In the context of visible light mediated photodeprotection, three strategies have primarily been used.
The first is direct deprotection via the excited state. Although this is challenging, exciting work is being done in this field, particularly with BODIPY dyes.5,6,10 The second is deprotection mediated through secondary thermal events. Recently, our group has had success in this area, developing a quinone trimethyl lock photocage.7 The third – a novel strategy that has been employed relatively recently – is the use of singlet oxygen to mediate oxidative events that lead to deprotection. This method has been met with significant excitement over the past few years9,11, however, the breadth of this strategy has yet to be explored. Of these three strategies, the third is most applicable to long wavelength photochemistry; dyes (such as indocyanine green) are known that absorb above 800 nm and generate singlet oxygen.12
Schnermann and coworkers demonstrated that irradiation of cyanine dyes leads to the formation of singlet oxygen which adds to the conjugated backbone (Figure 1).9 Following addition of singlet oxygen, subsequent collapse of the backbone occurs, which leaves behind an unstable intermediate. Further collapse of this intermediate can be coupled to ejection of a protecting group that will cyclize and release a free product. This process can be initiated across a range of cyanine dyes9,11,13,14, and can be modified to release amines. Despite the promise Figure 1: Proposed mechanism for cyanine
photodeprotection mediated by singlet oxygen.
of this system, it suffers from a number of flaws. The first is the speed. The process is inherently slow, due to both low singlet oxygen quantum yields inherent to cyanine dyes and subsequent dark steps that occur slowly. This inhibits its utility in dynamic systems – initial experiments required a 30-minute irradiation followed by a four-hour incubation at 37°C to achieve 60% photorelease.13 The second is the toxicity of the byproducts. Many of the cyanine dye decomposition byproducts are suspected to be toxic, limiting the use of the cage in biological systems. The final flaw is the substrate scope. Although recent papers have attempted to address this issue, the scope is still somewhat limited.9,11,14
4.2.2 New Approaches for Singlet Oxygen-Mediated Deprotection
To improve upon the concept of singlet oxygen mediated photorelease, a potent singlet oxygen generator must be combined with a singlet oxygen sensitive cage that is rapid and robust. For biological purposes the entire construct should be water soluble, suggesting an ideal starting point would be water soluble dyes. Although a number of water soluble dyes are known that generate singlet oxygen, phenothiazine dyes are particularly attractive, as they are synthetically accessible and can be modified for subsequent use. The most well-known phenothiazine dye is methylene blue (Figure 2A), which has found a number of uses in chemistry and biology. It is perhaps best known for its place on the World Health Organization’s List of Essential Medicines – as it is used to treat methemoglobinemia – however, from a photochemist’s perspective it is a robust singlet oxygen generator, with a quantum yield of 39%.15 Notably, methylene blue derivatives have been described in which the amino methyl groups are replaced with more complex alkyl groups (Figure 2B), providing a starting point for subsequent modification with singlet oxygen sensitive cages.16 Beyond methylene blue, some longer wavelength phenothiazine derivatives capable of generating singlet oxygen have been described, suggesting that singlet oxygen- mediated photodeprotection could be performed at wavelengths above 900 nm (Figure 2C).17,18
Although different singlet oxygen sensitive moieties have been described, two in particular standout.
The first is an alkene dithioether comprised of two thioethers bridged by an alkene (Figure 3A). Collapse of this group is initiated by the addition of singlet oxygen across the double bond, leading to formation
of a dioxetane. This dioxetane subsequently collapses to release thioesters, separating the previously linked components. This process is well characterized, and has been utilized in a number of applications where physical release is more important than the chemical context.19,20 This cage is not without its flaws – mainly lower yields due to thioether oxidation and a thioester scar that is appended to the leaving group – however, it is fairly robust and is a good starting point for initial explorations into singlet oxygen mediated photorelease. The second singlet oxygen sensitive moiety to consider is anthracene. Singlet oxygen is known to readily add across anthracene, which is actually the basis for a number of singlet oxygen detectors, such as singlet oxygen sensor green.21 Although this addition leads to no additional chemistry in the parent anthracene, in 9,10-dialkoxyanthracene derivatives (Figure 3B), the bridged oxygen will collapse following addition, leading to the formation of anthraquinone and the release of the alkoxy groups.22 This process has only been demonstrated using alcohols, however, it stands to reason Figure 2: Methylene Blue and associated phenothiazine derivatives. A.) Structure of methylene blue. B.) Structure of select methylene blue derivatives characterized by Gollmer et al.16 C.) Structure of NIR- phenothiazine derivative characterized by Hsieh et al.18
Figure 3: Singlet oxygen-mediated photorelease. A.) Alkene dithioether singlet oxygen-mediated photorelease. B.) Dialkoxyanthracene singlet oxygen mediated photorelease.
A
B
that similar reactivity would be obtained from anthracenes derivatized with amines and thiols, both of which are synthetically acessible.23,24
4.3 Results and Discussion
4.3.1 Exploration of Phenothiazine Derivatives as Singlet Oxygen Generators
To explore the possibility of singlet oxygen mediated-deprotection in the context of phenothiazines, derivatives were synthesized and evaluated for their ability to generate singlet oxygen. The first phenothiazine derivatives explored were based on a previously described long-wavelength phenothiazine derivative, o-DAP (Figure 2C). This dye has a λmax of 957 nm, and was shown to decrease the signal associated with a singlet oxygen sensor – 1,3-diphenylisobenzofuran (DPBF) – upon irradiation with a broad spectrum NIR light (700-1000nm), which suggests that the dye can form singlet oxygen.18 Despite the lack of a few notable controls, these results suggested that o-DAP was worth exploring further.
Although the literature provided a synthetic route that purported reasonable yields, it was found that the described synthesis was difficult to recapitulate, particularly due to side product formation via cross- coupling to the amine of dibromophenothiazine. An alternative path to the product was designed (Figure 4), one that also afforded modification of the terminal anilines prior to the final deprotection and oxidation steps. Briefly, phenothiazine was brominated in the presence of bromine and acetic acid to produce dibromophenothiazine. This compound was subsequently subjected to protection via Boc anhydride, which gave protected product in good yield. Other protection strategies were explored, but were less robust. Next, a Suzuki coupling was used to couple the chromophore-extending aniline groups.
Figure 4: Redesigned o-DAP synthesis. A new synthesis was designed to obtain a protected o-DAP precursor that could either be deprotected and oxidized to give o-DAP or modified prior to deprotection and oxidation give an o-DAP derivative. This synthesis was designed due to an inability to recapitulate a reported synthesis.
Due to the protection of the phenothiazine amine, this reaction proceeded efficiently, yielding a product that was ready for subsequent modification or deprotection/oxidation. Overall, the pathway to the protected product proved robust. More difficult, however, was determining conditions for subsequent deprotection and oxidation reactions.
Different deprotection conditions was explored, and ultimately 10% TFA in DCM was selected as the best deprotection condition. This left the product as a TFA salt, which could be quenched with saturated sodium bicarbonate and extracted into DCM. The obtained o-DAP precursor failed to exactly match the NMR spectra of the previously characterized o-DAP precursor despite having similar physical properties and the correct mass, suggesting that the deprotection left the dye in a slightly different state than the direct coupling. Subsequent oxidation of this dye with iodine led to a product with the UV-Vis (Figure 5) and mass spectra associated with o-DAP. However, an interpretable NMR could not be obtained, likely due to solubility issues. Different oxidation conditions were evaluated, however, iodine oxidation gave the best results.
Multiple attempts were made at producing singlet oxygen using a 940 nm LED and subsequently a 980 nm laser, but, in comparison to a dark control no singlet oxygen generation was detected by DPBF degradation (Figure 6). In fact, the dark control suggested that the published DPBF signal decrease was a consequence of some process that was occurring in the dark. Additional singlet oxygen sensors, such as 9,10-anthracenediyl-bis(methylene)dimalonic acid (ABDA), were tested to no avail. Taken together, this suggested that reported singlet oxygen generation by o-DAP was not going to be recapitulated.
However, before abandoning the dye completely, functionalization was explored to ensure that aggregation or solubility issues were not responsible for the observed results. Functionalization was initially attempted using the protected o-DAP precursor shown in Figure 6. Sulfonation was first evaluated
Figure 5: o-DAP UV-Vis Spectrum. Taken in DMSO.
using 1,3-propanesultone. Although product formation could be observed by LC-MS, the reaction mixture was a complicated milieu involving multiple sulfonation products, making it nearly impossible to cleanly isolate a single one. To further complicate matters, Boc deprotection was observed as the reaction proceeded. To avoid the complicated product mixture produced by 1,3-propanesultone, alkylation with other groups was next explored. Although the ultimate goal was alkylation with masked solubilizing groups – such as esters that could be deprotected to form carboxylic acids – methylation was explored as a proof of principle. Attempts to methylate the o-DAP precursor directly proved unsuccessful, however, a methylated precursor could be used for the coupling reaction, giving clean product. Upon deprotection and oxidation, this precursor gave tetramethyl o-DAP, which could be characterized by mass spectrometry and NMR (Figure 7). Although tetramethyl o-DAP was not soluble in water, it was soluble
Figure 7: Tetramethyl o-DAP synthesis.
Pre-Irradiation Post-Irradiation
Figure 6: Evaluation of o-DAP mediated singlet oxygen generation. Following irradiation with a 940 nm LED over the course of an hour, no significant change in DPBF signal was observed despite a notable decrease in o-DAP absorbance.
in chloroform, and therefore was irradiated in CDCl3 in the presence of DPBF in an attempt to generate singlet oxygen. Unfortunately, no significant change in DPBF signal was observed when compared to a dark control. This once again suggested that o-DAP was incapable of generating singlet oxygen. However, this successful methylation opened the door to the possibility of alkylation with other groups and prompted attempts to generate a water soluble o-DAP.
Alkylation of the protected o-DAP precursor with masked acetate groups was next evaluated using ethyl iodoacetate. Although this alkylation was cleaner and more robust than the previous o-DAP methylation attempts, it was still difficult to obtain complete alkylation. The use of long reaction times helped. Other alkylating reagents were also explored, such as methyl bromoacetate, however, only ethyl iodoacetate gave complete alkylation. Isolation of the alkylated o-DAP proved challenging, but clean product was obtained and verified by NMR (Figure 8). Subsequent Boc deprotection and oxidation reactions gave a chloroform soluble product with a reasonably clean NMR. This product still failed to convincingly produce singlet oxygen, demonstrating a decrease in DPBF signal in CDCl3 in both an irradiated sample and a dark control. Despite this failure, one final attempt was made to generate water soluble o-DAP in hopes that some solvent effect was preventing singlet oxygen generation.
Unfortunately, attempts to form carboxylic acids from the ester precursors were unsuccessful, generating a complicated mixture of products that could not be separated. In all, this suggested that o-DAP and its derivatives were not useful for generating singlet oxygen at near-infrared wavelengths. Before completely abandoning o-DAP as a singlet oxygen generator, a counterion exchange was attempted to ensure that the iodine counterion was not affecting the photochemistry. Exchange for tetrafluoroborate led to no change in the observed photochemistry.
Figure 8: Alkylated o-DAP synthesis.
Although the failure of o-DAP and its derivatives to produce singlet oxygen meant that long-wavelength photorelease was unlikely using phenothiazine derivatives, the accessibility of methylene blue derivatives suggested
that singlet oxygen-mediated photorelease should still be explored in the context of visible light. Although a number of phenothiazine derivatives were prepared according to a literature procedure16, a piperazine derivative was of particular interest as it contained an additional amine for further functionalization (Figure 9). Singlet oxygen sensitive cages were next explored.
4.3.2 Exploration of Singlet Oxygen Sensitive Cages As mentioned previously, two singlet oxygen sensitive cages were considered. Initial efforts were focused on the alkene dithioether cage,
specifically on functionalization of 1,2-dichloroethylene with alkane thiols, such as 2-mercaptoethanol (Figure 10). Ultimately these syntheses proved challenging, as complete separation of the alkane thiol precursor from the product was difficult to achieve. Although isolated cages could be cleaved in the presence of photochemically generated singlet oxygen, the process was far from robust, and focus was subsequently shifted to the development of anthracene photocages.
To explore anthracene photocages, an alcohol- terminal 9,10-dialkoxyanthracene was first synthesized (Figure 11). Although this dialkoxyanthracene was not water soluble – and therefore not ideal for biological applications – it provided an opportunity to explore a previously
described 9,10-dialkoxyanthracene synthesis. The synthesis proceeded as expected according to literature procedure, and the anthracene decomposed in the presence of photochemically generated singlet oxygen (as determined by HPLC). Anthraquinone was recovered following decomposition, demonstrating that Figure 10: 1,2-dichloroethylene functionalization with 2-mercaptoethanol.
Figure 9: Piperazine methylene blue derivative for derivatization.
Figure 11: 9,10-dialkoxyanthracene synthesis.
the photocage was working as previously described. These results suggested that heteroatom linked anthracenes were worth pursuing as singlet oxygen sensitive cages.
To explore the scope of anthracene photocages, the singlet oxygen-mediated release of heteroatoms other than oxygen was next evaluated. The incorporation of both amines and thiols at the 9 and 10 positions of anthracene has previously been described23,24, and initial synthetic attempts were based on literature precedent. Both dipropylamine and ethane thiol were successfully incorporated (Figure 12), generating a 9,10-diaminoanthracene and a 9,10-dithioanthracene respectively. To determine their sensitivity to singlet oxygen, each was combined with methylene blue and irradiated in methanol. Both anthracenes collapsed under these conditions and formed an identical product, ultimately identified to be anthraquinone (Figures 13 and 14). To ensure that this process was mediated by singlet oxygen, each sample was also irradiated after a freeze-pump-thaw procedure, which deoxygenated the solvent completely. The 9,10-dithioanthracene derivative was shown to be completely stable to irradiation of methylene blue under these conditions, suggesting that its decomposition was mediated solely by a path that required singlet oxygen formation. Interestingly, after freeze-pump-thawing, the 9,10- diaminoanthracene was still susceptible to decomposition upon irradiation in the presence of methylene blue, albeit at a significantly slower rate. This suggested that degradation of the 9,10-diaminoanthracene occurred via multiple pathways, of which not all required singlet oxygen formation. This could be due to electron transfer events mediated by methylene blue, a process that has previously been observed in
Figure 12: Synthesis of (A) diaminoanthracene and (B) dithioanthracene derivatives.
A
B
anilines.25 Regardless, the singlet oxygen mediated-pathway was still dominant in the 9,10- diaminoanthracene, as evidenced by the observed rate difference following freeze-pump-thawing. Taken together, these results demonstrated that 9,10-anthracene heteroatom derivatives were susceptible to degradation in the presence of singlet oxygen, which suggested that a large breadth of molecules could be released by singlet oxygen using anthracene cages.
Concurrent to explorations of heteroatom specificity, methods to increase the water solubility of 9,10- dialkoxyanthracenes were being explored. Initial attempts focused on increasing the solubility of the
Dark
Atmospheric, 30 min Irr FPT, 2 Hour Irr
Figure 14: 9,10-diaminoanthracene singlet oxygen sensitivity. Notably, degradation of the 9,10- diaminoanthracene occurs in the absence of oxygen (the FPT sample) suggesting that an oxygen-free photochemistry is occurring, albeit at a slower rate.
Dark Control
Atmospheric, 30 Min Irr
FPT, 1 Hour Irr
Figure 13: 9,10-dithioanthracene singlet oxygen sensitivity. Notably, degradation of the 9,10- dithioanthracene occurs only in the absence of oxygen.
caged group. The first route evaluated was selective oxidation of the terminal alcohols in 9,10- dialkoxyanthracenebispropanol. Although this oxidation did produce some terminally oxidized product, it was part of a complex product mixture that was too difficult to separate. The next route explored was generation of 9,10-dialkoxyanthracenes with esters that could be cleaved to give carboxylic acids.
Unfortunately, poor alkylation was observed, limiting progress along this synthetic route. Given the limited success of these attempts and the fact that the caged group itself is application dependent, subsequent attempts to increase solubility focused on modification of the anthracene cage.
2-chloroanthraquinone was chosen as a starting point for the development of more complex anthracene cages. Initial functionalization was attempted via condensation with piperazine, which gave 2-piperazinylanthraquinone upon heating in ethanol. This new cage had significantly increased water solubility, and attempts were made to generate a new 9,10-dialkoxyanthracene from this scaffold.
Unfortunately, subsequent attempts at reduction and alkylation were unsuccessful, likely due alkylation of the free amine. To avoid this side reaction, two avenues were explored. The first was to invert the order of modification, generating a 2-chloro-9,10-dialkoxyanthracene first and modifying it via condensation with piperazine second. Although a 2-chloro-9,10-dialkoxyanthracene was readily generated, it could not be subsequently modified to a 2-piperazinyl-9,10-dialkoxyanthracene. The second avenue explored was use of a protecting group to prevent the side reactivity associated with the piperazinyl amine. 1-Boc-piperazine was condensed onto 2-chloroanthraquinone, mimicking the piperazine condensation, however, this time the addition of a base (triethylamine) was required to give good yields. Subsequent alkylation to form a 9,10-dialkoxyanthracene was difficult to evaluate due to the insolubility of the product, but isolation of crude product and immediate Boc deprotection yielded a product that could be isolated. Although a combination of DCM/TFA gave better solubility for the crude
Figure 15: 2-piperazinyl-9,10-dialkoxyanthracene synthesis.
deprotection, an aqueous solution containing methanol and HCl was ultimately used as it gave cleaner conversion to product (Figure 15). The isolated 2-piperazine-9,10-dialkoxyanthracene was soluble in water and sensitive to photochemical singlet oxygen generation (Figure 16). Notably, when running an irradiation experiment using a freeze-pump-thawed sample, starting material disappearance was still observed, suggesting that piperazinyl modification of the anthracene cage enabled an oxygen free degradation mechanism not seen in the parent 9,10-dialkoxyanthracene. Given that this compound contains a pendant amine, it is likely that this molecule and 9,10-diaminoanthracene collapse by the same mechanism in the absence of oxygen.
4.3.3 Construction of a Scaffold for Singlet Oxygen Mediated Photorelease
Having demonstrated successful synthesis of a piperazinyl methylene blue derivative and 2- piperazinyl-9,10-dialkoxy anthracene (Figure 17), the final step was to combine the two components to generate a complete scaffold for singlet oxygen-mediated photorelease. To start with the simplest system possible, the addition of 2-piperazinyl-anthraquinone to oxidized phenothiazine was attempted, similar to the reaction used to generate piperazinyl methylene blue. The addition of an array of secondary amines
Dark Control
Atmospheric, 1 Hour Irr
FPT, 1 Hour Irr
Figure 16: 2-piperazinyl-9,10-dialkoxyanthracene singlet oxygen sensitivity. Notably, degradation of the 2-piperazinyl-9,10-dialkoxyanthracene occurs in the absence of oxygen, suggesting that oxygen-free photochemistry is occurring, albeit at a slower rate.